Question about Librem 14 and My RYF Idea

Hi. I ordered my Librem 14 last fall, shortly before the announcement that Librem 14s were going to move to firmware jails and Intel AX200 Wifi chips (Intel AX200 Wi-Fi/Bluetooth Shipping for New Orders – Purism).

I was generally happy with the fact that my Librem 14 arrived with an Atheros chip from before the move to Intel AX200. I am not trained in hardware, but the firmware jail sounds like gaming the system of what counts as RYF and what does not.

Generally I have really enjoyed my Librem 14. It has been good to me, and does many tasks generally well while also being fully free as far as I know. This seems very good – like the kind of good I was hoping for when I purchased it.

However, I have at times mentioned to friends when hanging out socially and using this laptop that maybe if I dropped this laptop off of a table and broke it, it is now irreplaceable. Because if I ordered a new one, the new one would have firmware jailed Intel AX200 stuff going on, which conceptually means that it’s probably not RYF in my head but also maybe still RYF because of the games played with definitions.

For example, if we read the above article, they tout the ability to choose which proprietary firmware blob to run instead of the one from the firmware jail as though this is an increased user freedom. I imagine that when Jonathon Hall wrote that post in that way, he really truly meant what he was writing. We could imagine there might be some government contractors who buy Purism devices in bulk and have access to the Intel wifi source code and therefore created their own advanced form of the proprietary wifi blob. Perhaps they run their advanced WiFi blob for an enhanced government security that I as a layman user do not have access to, for example.

So I am not disagreeing with what Jonathon Hall said. But rather, I am curious about the possibility of disagreeing with the definition of freedom to focus back more on words that I read on the FSF page in the past – namely, deciding that “a system choosing which black box binary blob to run” might be less free than a system which offers no such choice and seemingly has no need for such.

If I go by that definition, to the (very?) limited knowledge I have, it would seem like Atheros is better than Intel AX200. And thus, this feeling that I now have a pseudo-irreplaceable laptop.

But when I started to feel a little sad about that, I looked up the Atheros part number in my lspci, then dumped that number in some online web store that ships in my country, and the first thing that came up was a vendor claiming to sell similar Atheros chips for $10.

I did not do extensive research on the authenticity of whether those are really the same kind of Atheros chip, but I was curious if what I’m saying jives with the understandings of other users and if Atheros chips like the ones that used to come in Librem 14s might really sell for maybe $10 or $20, rather than being something no longer available?

Because, if that’s true, I don’t have to worry quite so much about ever dropping or damaging my Librem 14 because I could in theory buy another one then replace the WiFi chip with an Atheros myself. Is that the case?


Seems like it, but let us know if you end up buying one for yourself.

There are probably a couple of different scenarios.

  1. Being able to use a blob that is not in the jail allows you to, for example, try out a later version of the blob - in case it fixes some bug or adds some desired functionality - while being easily able to fall back on the jail copy if the trial doesn’t go well.
  2. Since Purism doesn’t want to tell you which WiFi card make and model you must run, there must always be some means of using a blob that is not the one that is in the jail, either by putting the alternative blob on the disk or replacing the blob in the jail. (It’s not as if Atheros and Intel need be the only choices for an M.2 WiFi card.)

At the end of the day, freedom includes the right to do privacy-unfriendly things like using random blobs.

I think the opinion expressed in the article is controversial, in the sense that there are differing opinions in this forum, basically relating to the question: Is an invisible blob better than a visible blob?

Note that we are talking about M.2 cards, not chips. Naturally it does matter what chip or chipset is on the card, in terms of what driver and/or blob might be needed, or indeed whether the card works at all under Linux.

If you had the choice between multiple cards that do not require the operating system to load a blob into the card then you are still choosing which blackbox blob to run - by choosing which card to install into your laptop.

If you really have no such choice (i.e. there is only one such card) then “Lack of choice is Freedom” sounds a bit like a slogan from 1984 rather than a realistic position. :wink:

My understanding is that the reason for moving away from Atheros is simply that reality had caught up with that card. It does not support anywhere near the latest WiFi standards. It’s a difficult sell when you have an up-to-date CPU with lots of cores, and lots of RAM and the latest NVMe drive … and the whole thing runs like a wet weekend because of the old WiFi.

For some customers WiFi may not be a requirement at all (that’s an option) or an old WiFi standard may be acceptable because the laptop is connected via ethernet much of the time or there’s just not a lot of network use.

The other risk is that eventually WiFi Access Points drop downward compatibility with older WiFi standards and your laptop simply can’t get on the WiFi at all. It is already bad to enable 802.11b support in your WAP. Eventually WAPs are just going to drop support for 802.11b, and likewise 802.11g and 802.11n will be dropped in time. If you need support for older WiFi standards then you would need to keep an older WAP - but eventually that WAP would die.

For $10 or $20, you tell us. :wink:

I would say that it is better to buy a couple of spares now because you never know when they really will become unobtainable. But, as said, without knowing what chip or chipset is on the card, it may be a waste of time, so I would ask the vendor - and you need to make sure that the firmware is just stored on the card since every WiFi card is going to need firmware from somewhere. In theory you should ask the vendor that too - but I can imagine not getting much comprehension of the question if put to many online vendors.

1 Like

I hear this argument more and more often, used especially to undermine the Free Software Foundation’s (FSF) Respects Your Freedom (RYF) standard specifying that firmware which is basically read-only and stored on a component can be thought of as hardware, so it does not disqualify a system from RYF certification.

To me, the RYF standard makes sense, so it is frustrating that so many people try to undermine it. I believe many people making this argument are sincere and that many are disingenuous shills for the many interests who stand to profit from either the FSF relaxing its standard or people not valuing RYF certification.

Based on years of interacting with you on these forums, @irvinewade, I trust you are not a shill, so in response to you making this argument, I began to think, how can I push this issue forward in a productive way? To that purpose, maybe I can clarify some positions:

First of all, what you said above is true, if by “freedom” you mean a particular person’s freedom to make a particular choice between different binary blobs. So, yes, more freedom in that particular way.

However, even though operating system (OS)-loaded blobs offer more of that kind of freedom, I would say they are less Free Software than component-flashed binary blobs.

Now, you might say, “weirdnerd, that is not true! The OS-loaded blob is more free, because it allows the freedom to modify, one of the four software freedoms! The component-flashed blob doesn’t allow that!”

To that, I will say the OS-loaded blob imperfectly allows the freedom to modify, because to effectively modify a program requires access to the program’s source code. Nevertheless, I will admit that the OS-loaded blob is more free.

“WHAT?!? Weirdnerd, you just said above that the OS-loaded blob is less Free Software! Now you admit the opposite!”

No, I admit it is more free, but I maintain it is less Free Software. Not because it is less free, but because it is less software! It’s even in the name firmware :slight_smile: The component-flashed blob is more firm than the OS-loaded blob, meaning the OS-loaded blob is more soft. Software versus firmware versus hardware.

So, finally, in answer to your comment above, yes, OS-loaded blobs allow more of that particular kind of freedom, but they undermine the furtherance of Free Software.

I will save for another comment my argument that furthering Free Software along those lines is a good thing, but I believe that it is.

@Dlonk, I will be curious to see if you can successfully ensure a fully Free Software L14 (firmware jail caveats included) for yourself for the future. Or maybe someone else can. Good luck!


It used to be common that when devices went out of production that discounters would buy up leftover inventory and resell at low prices. This happens less often now, but still happens from time to time.

wifi cards can and do die, so having a spare might be handy if you think an acceptable replacement won’t be available new. On ther other hand, if your laptop dies early because of accident, the wifi card might survive and could be installed on the replacement laptop.

1 Like

That model atheros chipset on a standard M.2 should be inexpensive if available. At some point they will probably stop being produced since they never worked all that well and they don’t support the newer Wifi standards. Part of why they didn’t work well is that they don’t actually have loadable firmware (true RYF compliance). That firmware is very basic … and they pushed the complication into the driver. There are just some issues that haven’t been dealt with using the FOSS driver.

See ath9k and ath9k_htc under Comparison of open-source wireless drivers - Wikipedia for more info.

Also, one can hold out hope that the list of wifi cards with FOSS drivers + firmware expands.


My comment was intended to address what scenarios are available to the OP in respect of his Librem 14 rather than to address anything about RYF. So I didn’t mention RYF at all other than implicitly to make the side observation that

I think the opinion expressed in the article is controversial, in the sense that there are differing opinions in this forum, basically relating to the question: Is an invisible blob better than a visible blob?

My own opinion is that blobs are bad, blobs are fundamentally limiting of your freedom, blobs undermine the value of your investment in the hardware. The location of the blob is a secondary consideration.

There is also a difference between a component that stores the blob inside the component and provides no means to update the blob and one that does provide some means of updating the blob. (Annoyingly, it is not unheard of that the only way of updating the internal blob is from Microsoft Windows. So to update one small internal blob requires the use of one giant blob.)

That may be true but, as I said, I think reality just caught up with the original M.2 card that Purism was using. In other words, in my opinion, you have to pick your battles.

I would guess that Purism doesn’t want to maintain inventory of the old card now, or to have to offer 3 choices at order time (no card, new card, old card) - so the option right now is to order with no card, source your own old card, and away you go.


A post was merged into an existing topic: Does Purism Plan on Adding 5G as an Option for the Librem 5?

You are the best!

Purism | Free Software :gem: